Scribd Logo
Upload

Welcome to Scribd!

  • Tilings by Regular Polygons

    Uploaded by

    kiwigirl74
    12 views21 pages
    Patterns in the plane from Keplerto the present, including recent results and unsolved problems.

    Copyright

    © © All Rights Reserved

    Available Formats

    PDF or read online from Scribd

    Did you find this document useful?

    Is this content inappropriate?

    Report this Document
    Patterns in the plane from Keplerto the present, including recent results and unsolved problems.

    Copyright:

    © All Rights Reserved
    Download as PDF or read online from Scribd
    Flag for inappropriate content
    12 views21 pages

    Tilings by Regular Polygons

    Uploaded by

    kiwigirl74
    Patterns in the plane from Keplerto the present, including recent results and unsolved problems.

    Copyright:

    © All Rights Reserved
    Download as PDF or read online from Scribd
    Flag for inappropriate content
    of 21
    AZuUCLES Tilings by Regular Polygons Patterns in the plane from Kepler to the present, including recent results and unsolved problems. Branko Gronsaum University of Washington Seattle, WA 98195 Grorrrey C. SHEPHARD University of British Columbia Vancouver, British Columbia, VT 1WS A fling of the plane is a family of sets — called tiles — that cover the plane without gaps or overlaps. (“Without overlaps” means that the intersection of any two of the sets has measure (area) zero.) Tilings are also known as tessellations, pavings, or mosaics; they have appeared in human activities since prehistoric times. Their mathematical theory is mostly elementary, but nevertheless it contains a rich supply of interesting and sometimes surprising facts as well as many challenging problems at various levels. The same is true for the special class of tilings that will be discussed here — ‘more or less regular tilings by regular polygons. These types were chosen because they are accessible without any need for lengthy introductions, and also because they were the first to be the subject of mathematical research. The pioneering investigation was done by Johannes Kepler, more than three and a half centuries ago. Additional historical data will be given later (in Section 6) but as an introduction we reproduce in Ficure 1 certain drawings from Kepler [1619]. We shall see that these drawings contain (at least in embryonic form) many aspects of tilings by regular polygons which even at present are not completely developed. As is the case with many other notions, the concept of “‘more or less regular” tilings by regular polygons developed through the centuries in response to the interests of various investigators; itis still changing, and no single point of view can claim absolute superiority over all others. Our presentation reflects our preferences, although many other definitions and directions are possible; some of these will be briefly indicated in Sections 4, 5 and 7. For most of our assertions we provide only hints which we hope will be sufficient for interested readers to construct complete proofs. Initially we shall use only regular convex polygons as tiles: if such a polygon has n edges (or sides) we shall call it an n-gon, and use for it the symbol {n}. Thus {3} denotes an equilateral triangle, while {4}, {5}, {6} denote a square, a (regular) pentagon, and a (regular) hexagon, respectively. All the polygons are understood to be closed sets, that is, to include their edges and vertices. VOL. 50, NO. 5, NOVEMBER 1977 227 Various more or less regular tilings of the plane by regular polygons, reproduced from J. Kepler’s book “Harmonices Mundi”, published in 1619. Fiore 1 Except in Section 4 we shall restrict attention to tilings that are edge-to-edge; by this we mean that as far as the mutual relation of any two tiles is concerned there are just three possi i) they are disjoint (have no point in common); (ii) they have precisely one common point which is a vertex of each of the polygons; or (iii) they share a segment that is an edge of each of the two polygons. Hence a point of the plane that is a vertex of one of the polygons in an edge-to-edge tiling is also a vertex of every other polygon to which it belongs; we shall say that itis a vertex of the tiling. Similarly, each edge of one of the polygons is an edge of precisely one other polygon and we call it an edge of the tiling. 1, Regular and uniform tilings The question about the possibilities of tiling the plane by (congruent) copies of a single regular polygon has the following simple and rather obvious answer, the origin of which is lost in antiquity The only possible edge-to-edge tilings of the plane by mutually congruent regular convex polygons are the three regular tilings by equilateral triangles, by squares, or by regular hexagons. A portion of each of these three tilings is illustrated in Ficure 2. 228 MATHEMATICS MAGAZINE yyy vt ———— . : x COOP Yrrr Ee OOM CO x / KA TI - VAVAVAVAVAVAVAVAN TT POO FO Vo \ VN ey a) @) ‘The three regular tilings of the plane. Ficure 2 If we inquire about the possibility of edge-to-edge tilings of the plane that use as tiles regular polygons of several kinds, then the situation immediately becomes much more interesting. The angle at each vertex of {1} is (1 — 2)/n so it is easy to check by simple arithmetic that only 17 choices of polygons can be fitted around a single vertex so as to cover a neighborhood of the vertex without gaps or overlaps. We call each such choice the species of the vertex, and list in Taste 1 the 17 possible species. In four of the species there are two distinct ways in which the polygons in question may be arranged around a vertex; the mere reversal of cyclic order is not counted as distinct. Hence there are 21 possible types of vertices; they too are listed in TABLE 1 and also illustrated in Figure 3. We denote the type of a vertex around which there are, in cyclic order, an a-gon {a}, a b-gon {b}, a c-gon {c}, etc., by a.b.c...... Thus the three regular tilings have vertices of types 3.3.3.3.3.3,4.4.4.4, and 6.6.6. For brevity we shall write these symbols as 3°,4* and 6°, and we shall use similar abbreviations in other cases. In order to obtain a unique symbol for each type of vertex we shall always choose that which is lexicographically first among all possible expressions. Contrary to frequently made assertions (see Section 6), if we require of an edge-to-edge tiling only that it be composed of regular polygons and that all its vertices be of the same species, then there are infinitely many distinct types of tilings. For example (see FicuRe 4), if at each vertex there are two triangles and two hexagons, it is possible to place each “horizontal” strip in two non-equivalent x xx X By sliding horizontal strips independently of each other, an uncountable infinity of distint tilings may be obtained, all vertices of which are of species 5. Figure 4 VOL. 50, NO. 5, NOVEMBER 1977 229 3.8.24 4444 3.4.3.12 ~ / 3.9.18 3.3.4.12 3.12.12 4.5.20 4.6.12 3.10.15 ‘The 21 possible types of vertices. Figure 3 230 MATHEMATICS MAGAZINE positions. Since there are infinitely many such strips, there will be uncountably many distinct tilings. The situation in Fiaure 5 in which 3 triangles meet 2 squares at each vertex is similar. In FIGURE 6 we allow three kinds of polygons; this permits each “disc” of an infinite family to be put in two positions, again leading to uncountably many tilings, with each vertex of species 6. In view of the above remarks it is reasonable to restrict attention to tilings in which only a single type of vertex is allowed. If that type is a. b.c. ..., we shall denote the tiling by (a..b.c. ...), using superscripts to shorten the expression when possible. This restriction indeed changes the situation completely and we have the following result: There exist precisely 11 distinct types of edge-to-edge tilings by regular polygons such that all vertices of the tiling are of the same type. These 11 types of Species 4 5 6 7 8 9 1 12 15 18 2 2% 42 Typeot — Typeof number vertex tiling 1 6 3.3.3.3.3.3 4 2 4 1 3.3.3.3.6 A 33 (2 3.3.3.44 A 3343.4 A 420 1 3.34.12 3.4.3.12 52 2 3.3.6.6 3.6.3.6 A 6 12 1 3.4.4.6 3464 A 7 4 1 13.7.8 so 1 1 3.8.24 9 1 1 1 3.9.18 0 1 1 3.10.15 no 2 3A 2 4 4444 A 1B 1 1 4.5.20 14 1 1 1 4.6.12 A 15 1 2 488 A 16 2 1 5.5.10 "7 3 6.6.6 A POSSIBLE SPECIES AND TYPES of vertices for edge-to-edge tilings by regular polygons. Entries in the table indicate the number of n-gons that meet at a vertex. Types that lead to Archimedean tilings are labelled with an “A” {in the final column. TapLe 1 VOL. 50, NO, 5, NOVEMBER 1977 231 Infinitely many distinct tilings that have only vertices of species 3 may be obtained by changing the relative positions of horizontal zigzag strips in the tiling at the left. Figure $ tilings, illustrated in Figure 7, are usually called Archimedean tilings (although some authors call them homogeneous, or semiregular, or uniform). They clearly include the three types of regular tilings Two not entirely trivial steps are required in order to prove that there are precisely 11 types of Archimedean tilings. In the first place, it must be shown that for ten of the 21 types of vertices listed in Taaue 1 it is not possible to extend a tiling from the neighborhood of a starting vertex to an Archimedean tiling of the whole plane. In fact, in each case one has to go only around one of the rn-gons with odd n to show the impossibility. (For each of the six species numbered 7, 8, 9, 10 13 and 16 there is no edge-to-edge tiling of the plane by regular polygons that includes even a single vertex of the species.) In the second place, it must be established that the remaining 11 types of vertices do actually lead to Archimedean tilings. This may be deemed obvious and trivial in view of Fiouze 7, but it is just this “obviousness” that is dangerous. In Ficure 8, adapted from a children’s coloring book, wwe show a tiling that appears to consist of regular n-gons with n = 4,5,6,7,8, Actually, this visual “proof” is a fraud, since it is easy to check that the polygons in such a tiling cannot be regular. Thus there is a real need to show that the 11 Archimedean tilings do exist. It is easy to give direct proofs of existence for (4*) and for (3°) by considering two or three suitable families of equidistant parallel lines. The existence of the other Archimedean tilings can be deduced (with just a little thought) from these two. ~ NO \ aT_ IN TT By turning “discs” in the tiling at the left infinitely many different obtained. lings with all vertices of species 6 may be Fioure 6 232 MATHEMATICS MAGAZINE eB) ay 6) be K 4 DKK PB ~eSes< 4 POPS - ae: T he T 64.6) Ge) (7.4.3.4) G.4.6.4) 6.6.3.6) G12) ‘The 11 stint types of Archime- dean ings ofthe plane. The ting of type (3*. 6) exists in two mirror- —* symmetric (enantiomorphic) forms. FouRe 7 6.2) ae) VOL. 50, NO. 5, NOVEMBER 1977 233, It should be noted that a prior’ it is not obvious that the limitation to a single type of vertex should lead to a single type of tiling. It happens to turn out that way, but just barely so: it is only because we find it convenient not to distinguish between tilings that are congruent but not directly congruent. Indeed, the tilings of type (3*.6) are of two mirror-symmetric (enantiomorphic) forms that are counted as distinct by some authors. Another accidental but very important feature of the Archimedean tilings is the fact that each is vertex-transitive. By this we mean that all vertices are equivalent under the symmetries of the tiling. Put more simply, for each pair of vertices A and B it is possible to find a motion of the plane, or a motion combined with a reflection in a line, that carries the tiling onto itself and maps A onto B. A verification of the vertex-transitivity of the Archimedean tilings is a very useful exercise. A psychologically very convincing (although logically not completely conclusive) verification of the transitivity may be obtained by tracing the tiling on a transparent sheet that may be moved over the original, and turned over. (Note that a tiling may be vertex-transitive even if its tiles are not regular polygons. Some examples of such tilings will be found in FiGurés 14 snd 16.) In view of the transitivity of Archimedean tilings we shall from now on also call them uniform tilings. The distinction between the two words is that “Archimedean” refers only to the fact that the immediate neighborhoods of any two vertices “look the same”, while the term “uniform” implies the much stronger property of equivalence of vertices under symmetries of the whole tiling. Returning to the question of tiling with a single species of vertex we mention without proof that non-uniform tilings are possible only in case of species 3, 5 and 6. In the last two of those cases all tilings can be obtained from the uniform ones, (3.6.3.6) and (3.4.6.4), by the method explained above. However, in case of species 3 there are other possibilities as well and a complete description of all such tilings is still not known. In the same vein, it may easily be verified that the three regular tilings have the following strong property. If a triplet consisting of a polygon, one of its edges, and a vertex of that edge is called a flag, then any two flags of a regular tiling are equivalent under the symmetries of the tiling. From now on, the “regularity” of “regular tilings” will always be understood in this sense, which is becoming more widespread in many related areas; see, for example, Coxeter [1975}, Griinbaum [1976]. We should stress that flag-transitivity is more restrictive than requiring that a tiling be vertex-, edge- and tile-transitive: there is exactly one tiling by polygons (Ficuke 16a) which has the latter three kinds of transitivity, but which fails to be regula 2. keuniform tilings The observation that the Archimedean tilings are uniform suggests the following possibility of generalization, A tiling is called k-uniform if its vertices form precisely k transitivity classes with respect to the group of all symmetries of the tiling. Clearly, uniform tilings coincide-with I-uniform tilings. If the types of vertices in the k classes are ayy C1. ...3 42.2. €2-+2+3++45 ks Deo Cue +15 We will designate the tiling by the symbol (a).b;.cy....3@2-b2-C2- jdy.by.e....), with the obvious shortening through the use of superscripts, and with subscripts to distinguish tilings in which the same types of vertices appear. There exist 20 distinct types of 2-uniform edge-to-edge tilings by regular polygons. They are shown in Figure 9. The proof of this fact may be carried out along lines analogous to those explained in connection with the 11 uniform tilings. However, the details are here much more intricate; it appears that the only published version of the proof is found in the paper of Kritenheerdt [1969]. It is not hard to see that there exist k-uniform tilings for each k 21. Examples are given in Krdtenheerdt (1969] for k = 3,4,5,6 and 7, and in Ficure 10 for k = 3,4. However, even for k = 3itis not known how many distinct 3-uniform tilings exist, nor is any kind of asymptotic estimate available for the number of k-uniform tilings with large k A closely related notion was also examined by Krétenheerdt [1969], [1970], [1970b]. He considered those k-uniform tilings in which the k transitivity classes of vertices consist of k distinct 234 MATHEMATICS MAGAZINE SDAL PAA KROL “EEE B53".6), B%3".6)s (53.4), B53".4) , Ne @38.4.3.4) 3.4.12) 437.6)" 53.6) AU a 3 o { i 4 a6 Se 71 J O4537.4.3.4)y @.4;3.4.6.4) Ca ayt ‘ Y t A C454, .4.3.453.4.6.4) .653.6.3.6" G.4.3.12;3.123° ALi yy aoe AA : UW aH mI J L 7 4 J (4 - S Prt Pt 6.#.6;:3.4.6.4) G.4.6;3.6.3-6), 6.4 .6;3.6.3.65 G.4.6.4;4.6.12)° ‘The 20 different types of 2-uniform tilings. The tiling (3; 3*. 6), exists in mirror-symmetric forms, only one of which is shown. Tilings marked by an asterisk are homogeneous in the sense defined in Section 3. One vertex of each transitivity class is marked. Fioure 9 VOL. 50, NO. 5, NOVEMBER 1977 235 AA fake tiling with regular polygons, adapted from a children’s coloring ‘Two examples of homogeneous tilings; one is also 3-uniform, the other book Aliair Design (Holiday {s 4-uniform, One vertex of each transitivity class is marked. ts7p. Fioure 8 Figure 10 types of vertices. While itis easily seen that for k = 1 and for k = 2 these coincide with the k-uniform ones, Krétenheerdt’s condition is actually restrictive for k =3. Denoting by K(k) the number of distinct Krotenheerdt tilings, he established that K(1) = 11, K(2)= 20, K@)=39, K(4) = 33, K(S)= 15, K(6)=10, K(7)=7 and K(k)=0 for each k 28. Krotenheerdt’s method of proof is a natural extension of the one used in the determination of the uniform tilings 3. Homogeneous and edge-transitive tilings Departing from the terminology used by some authors, we shall say that an edge-to-edge tiling of the plane by regular polygons is k-homogeneous if the tiles form precisely k transitivity classes under the symmetries of the tiling. We shall also say that a tiling is homogeneous if all tiles that are mutually congruent form one transitivity class. It is easily verified that all the uniform tilings are homogeneous, except (3*.6), which is 3-homogeneous. Other homogeneous tilings are the seven 2-uniform tilings marked by an asterisk in FicuRe 9 Itis rather surprising that there seems to be no consideration in the literature of the homogeneous or k-homogeneous tilings. It appears reasonable to expect that for each k there exists a least number ‘h(k) such that every k-uniform tiling is also h-homogeneous for some h $h(k). Likewise, there probably exists a least number k(h) such that every h-homogeneous tiling is also k-uniform for some k Sk(h). From the above remarks and from FiGure 9 it is easy to see that h(1)=3 and h (2) = 5. On the other hand clearly k(1)=1, while the examples of Ficure 10 show that k(2)2 4, ‘The determination of all 2-homogeneous tilings (all of which are, obviously, homogeneous) should not be very hard, and even the determination of all homogeneous tilings is probably possible with a little patience, We conjecture that the 19 homogeneous tilings shown in Ficures 7, 9 and 10 are the only ones possible, and that, in consequence, there are just fourteen 2-homogeneous tilings, and that k(2)=4. Similar problems arise if we consider transitivity classes of edges. If there are j such classes in a tiling we shall call it a j-edge-transitive tiling. We mention this idea because we believe that it also is not considered in the literature. There appear to be just four I-edge-transitive tilings by regular polygons (namely (3°, (4), (6°) and (3.6.3.6)) and four 2-edge-transitive tilings (namely (3.4.3.4), 3.4.6.4), (3.12) and (4.8%). 4, Tilings that are not edge-to-edge We now consider tilings by regular polygons without the requirement that the tiles meet edge-to-edge. Kepler briefly considered this possibility (see drawings Bb and Kk in Ficuze 1), but no further consideration seems to have been given to the mathematical possibilities for several centuries. 236 MATHEMATICS MAGAZINE 0

    You might also like